Player evaluation with tracking data in sports analytics

Establish a comprehensive and rigorous methodology for player evaluation using player tracking data in sports analytics, enabling assessment of players based on their observed positioning and movement captured throughout games.

Background

The paper situates its contributions within the broader context of sports analytics, where high-frequency player tracking data have enabled detailed modeling of within-play dynamics. Despite advances, the authors emphasize that a general, principled solution for evaluating players directly from tracking data has not been definitively established.

To address a portion of this broader need, the authors develop Bayesian multilevel step-and-turn models to evaluate frame-level movement of NFL ball carriers by comparing observed actions to posterior predictive distributions of hypothetical alternatives. While their contribution advances the state of the art in a specific setting, the overarching problem of player evaluation using tracking data across sports and contexts is highlighted as remaining open.